In my mind one of the key things (besides visualization) Sourcetrail provides is a useful graph structure for source code that can be used for various types of analysis, without having to go through the pain of writing a clang analyzer yourself. However, since sourcetrail doesn't have plugins (#1076), writing your own analysis tools is a bit of work, and may not let you use other kinds of graph an

The ScanDirectory method of FileScanner allocates pointers with new and returns raw pointers. It should use unique_ptr instead.

Note: Wait for #13444 resolution before working on this.

Current implementation of join can be improved by performing the operation in a single call to the backend kernel instead of multiple calls.

This is a fairly easy kernel and may be a good issue for someone getting to know CUDA/ArrayFire internals. Ping me if you want additional info.

Hi. I can imagine calendar options that would allow people to try different types of calendars. For example with lunar calendars like the Chinese, Vietnamese, Hindu, and Thai calendars.
I see this a powerful aid in teaching the importance of calendars and how they can help us relate to cosmic influences.
By the way... stellar work so far!

It just doesn't test anything for the cast majority of fuzzer input. While fuzzer does learn the effective paths, let's help it more.

First, for instruction type, change the switch argument to: instruction.operation % NUM_OPERATIONS.

Then just split the available values into the two sets by looking up one integer from fuzzer and use that module size for set1 and the rest for set2.

In order to support package managers better, which write their retrieved data typically in unique paths, it would be nice to be able to declare multiple base_dir paths to translate absolute paths in relative ones.

For example a package manager would have the following paths in its cache:

~/data/libA/1.0.0/acme/stable/package/63da998e3642b50bee33f4449826b2d623661505/include/libA/myheader